A red meat-derived glycan promotes inflammation and

cancer progression
Annie N. Samraja,1, Oliver M. T. Pearcea,1, Heinz Lublia, Alyssa N. Crittendena, Anne K. Bergfelda, Kalyan Bandaa,
Christopher J. Gregga, Andrea E. Bingmana, Patrick Secresta, Sandra L. Diaza, Nissi M. Varkia,b, and Ajit Varkia,c,2
Departments of cMedicine, bPathology, and aCellular and Molecular Medicine, Glycobiology Research and Training Center, University of California, San Diego,
La Jolla, CA 92093

A well known, epidemiologically reproducible risk factor for human

carcinomas is the long-term consumption of red meat of mammalian origin. Although multiple theories have attempted to explain
this human-specific association, none have been conclusively
proven. We used an improved method to survey common foods
for free and glycosidically bound forms of the nonhuman sialic acid
N-glycolylneuraminic acid (Neu5Gc), showing that it is highly and
selectively enriched in red meat. The bound form of Neu5Gc is bioavailable, undergoing metabolic incorporation into human tissues,
despite being a foreign antigen. Interactions of this antigen with
circulating anti-Neu5Gc antibodies could potentially incite inflammation. Indeed, when human-like Neu5Gc-deficient mice were fed
bioavailable Neu5Gc and challenged with anti-Neu5Gc antibodies,
they developed evidence of systemic inflammation. Such mice are
already prone to develop occasional tumors of the liver, an organ
that can incorporate dietary Neu5Gc. Neu5Gc-deficient mice immunized against Neu5Gc and fed bioavailable Neu5Gc developed
a much higher incidence of hepatocellular carcinomas, with evidence of Neu5Gc accumulation. Taken together, our data provide
an unusual mechanistic explanation for the epidemiological association between red meat consumption and carcinoma risk. This
mechanism might also contribute to other chronic inflammatory
processes epidemiologically associated with red meat consumption.

red meat and cancer N-glycolylneuraminic acid

tumor-associated inflammation tumor-associated carbohydrate antigen
xenosialitis

exposure (15). Overall, although more than one of these theories

may still prove to be correct, definitive proof remains lacking.
Another unexplained fact is the human specificity of this risk
(i.e., other vertebrate carnivores do not suffer a high incidence of
carcinomas). In this regard, we have suggested an unusual
human-specific mechanism, involving inflammation associated with
metabolic incorporation of a nonhuman sialic acid, N-glycolylneuraminic acid (Neu5Gc), and interaction with circulating
anti-Neu5Gc antibodies (1619). Despite the fact that humans
are genetically unable to produce Neu5Gc, this molecule is detectable on surfaces of human epithelia and endothelia, and in
higher amounts in malignant tissues (20). In the absence of an
alternate pathway for Neu5Gc biosynthesis (21), the only possible source for incorporation is dietary intake (22). An initial food
survey showed a prominent presence of Neu5Gc in red meat
(23). Metabolic incorporation of dietary Neu5Gc into tissues
(24) makes this glycan the first example, to our knowledge, of a
xeno-autoantigen, which can react with circulating anti-Neu5Gc
antibodies (i.e., xeno-autoantibodies) (25). The resulting antigenantibody interaction is hypothesized to generate or promote
chronic inflammation or xenosialitis, which could contribute
to carcinogenesis or to other diseases exacerbated by chronic
inflammation.

Significance
We present an unusual mechanism for the well-known association between red meat consumption and carcinoma risk involving the nonhuman sialic acid N-glycolylneuraminic acid
(Neu5Gc). We first evaluate the Neu5Gc content of various
foods to show that red meats are particularly rich in orally
bioavailable Neu5Gc and then investigate human-like Neu5Gcdeficient mice fed this form of Neu5Gc. When such mice were
challenged with anti-Neu5Gc antibodies, they developed evidence of systemic inflammation. Long-term exposure to this
combination resulted in a significantly higher incidence of
carcinomas (five-fold increase) and an association with Neu5Gc
accumulation in the tumors. Similar mechanisms may contribute to the association of red meat consumption with other
diseases, such as atherosclerosis and type 2 diabetes, which are
also exacerbated by inflammation.

here is a long-standing epidemiological link between the

consumption of red meat (beef, pork, and lamb) and the
incidence of carcinomas, atherosclerosis, type 2 diabetes, and allcause mortality (14). Although such diseases have multifactorial
origins, all are aggravated by chronic inflammation (5, 6). Red
meat-rich diets also correlate with circulating markers of inflammation and endothelial dysfunction (7). Here, we focus on red
meat-related risk of carcinomas (further citations regarding the
association are provided in Table S1). Corroboration comes from
the low rates of carcinomas in populations that consume very low
levels or no red meat (810). Within the World Cancer Research
Foundation report, red meat was among the top 10 factors associated with incidence and progression of carcinomas in all populations (11). Enhancement of carcinoma risk appears to be highest
in tissues like colonic epithelium, in which adenomas can progress
to carcinomas, driven by molecular changes in oncogenes and tumor suppressor genes (12). There are many proposed mechanisms
for the cancer-promoting effects of red meat (13), including generation of mutagens by grilling, DNA damage due to N-nitroso
compounds, or free radical generation by heme iron. However,
none of these mechanisms have been proven, and confounding facts
are apparent in some instances (e.g., grilling of poultry and fish
generates the same mutagens, yet these foods are not associated
with cancer risk) (14). Also, doses of the mutagens that induce
carcinomas in animal models are many fold higher than human
www.pnas.org/cgi/doi/10.1073/pnas.1417508112

Author contributions: A.N.S., O.M.T.P., A.N.C., K.B., N.M.V., and A.V. designed research;
A.N.S., O.M.T.P., H.L., A.N.C., A.K.B., K.B., C.J.G., A.E.B., P.S., S.L.D., and N.M.V. performed
research; A.N.S., O.M.T.P., H.L., and A.K.B. analyzed data; and A.N.S., O.M.T.P., H.L., and
A.V. wrote the paper.
Conflict of interest statement: A.V. and N.M.V. are cofounders of and advisors to SiaMab
Therapeutics, Inc., which has licensed University of California, San Diego technologies related
to anti-Neu5Gc antibodies in cancer.
This article is a PNAS Direct Submission.
1

A.N.S. and O.M.T.P. contributed equally to this work.

To whom correspondence should be addressed. Email: a1varki@ucsd.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.

1073/pnas.1417508112/-/DCSupplemental.

PNAS Early Edition | 1 of 6

MEDICAL SCIENCES

Edited by Stuart A. Kornfeld, Washington University School of Medicine, St. Louis, MO, and approved November 25, 2014 (received for review September
18, 2014)

Standard Gc

Dairy
(Cows Milk)

Ac

Ac

Gc

Red meat
(Beef) Gc
Ac

Fluorescence intensity

Although attractive in principle, this hypothesis has not been

proven in an in vivo system. Notably, Cmah/ (CMP-Neu5Ac
hydroxylase knockout) mice with a human-like deficiency of
Neu5Gc (24) can incorporate Neu5Gc of dietary origin, but only if
provided in the bioavailable glycosidically bound form (22). However, such mice do not spontaneously generate anti-Neu5Gc
antibodies. The likely reason is that humans are spontaneously
immunized via an unusual postnatal process involving a humanspecific commensal bacterium that presents Neu5Gc from
weaning foods to the immune system (26). However, an antiNeu5Gc response is induced in Cmah/ mice by active immunization, and specific antisera can be generated. Using such
antisera in Cmah/ mice bearing syngeneic Cmah+/+ tumors, we
have shown that the anti-Neu5Gc antibodies could facilitate
tumor progression by enhancing inflammation (16, 17). However, despite this suggestive evidence, we have not demonstrated
that oral feeding of Neu5Gc can induce the proposed xenosialitis
in vivo or that this process can increase rates of spontaneous
carcinomas. Also, dietary surveys on Neu5Gc in food have been
limited, the ratio of bound and free Neu5Gc has not been determined, and effects of food processing have not been considered. Given our hypothesis that tissue incorporation of Neu5Gc
of dietary origin can have deleterious effects, it is important to
determine the amounts present in typical components of the
Western diet accurately. Although earlier studies attempted to
quantify Neu5Gc distribution in food (23, 2729), a systematic
catalog of a wide variety of commonly consumed foods is lacking.
Furthermore, based on recent studies (22), it appears that glycosidically bound Neu5Gc is the dietary source that is bioavailable for tissue incorporation, and not the free monosaccharide.
However, available data neither differentiated between bound
and free Neu5Gc nor addressed the effects of food processing.
Here, we address all these issues by first developing an assay to
determine amounts of free and bound Neu5Gc accurately in a wide
variety of foods. We then evaluate the impact of short-term and
long-term feeding of Neu5Gc in Cmah/ mice combined with
passive or active immunization against Neu5Gc. Although the association between red meat consumption and colon cancer is most
prominent in humans, we focus here on hepatocellular cancer in
male C57BL/6 mice as proof of principle, because these animals
have a low rate of spontaneously occurring hepatic tumors (30).

Poultry
Ac
(Chicken Egg)

Seafood
(Salmon)

Ac

Vegetable
(Carrot)

Results
Distribution of Free and Bound Neu5Gc and N-Acetylneuraminic Acid
in Foods. Acid hydrolysis and derivatization by 1,2-diamino-4,5-

methylenedioxybenzene dihydrochloride (DMB) followed by

RP-HPLC with fluorescent detection (DMB-HPLC) has been
the method of choice for the detection and quantification of
sialic acids (31). Nevertheless, this method cannot accurately
differentiate between bound and free sialic acids, because the
conditions of the derivatization reaction (1.5 M acetic acid,
50 C, 2.5 h) result in some concurrent release of bound sialic
acid (this false-positive signal varies with the sample; 8.6% for
a model substrate, such as fetuin; Table S2). For the current
study, we therefore modified conventional protocols by adapting
temperature derivatization conditions (4 C for 48 h) originally
introduced for preserving acid-labile polysialic acid (32). These
conditions efficiently tag free sialic acids, without any significant
release of bound sialic acids (Table S2). By studying a parallel
aliquot with the conventional method, we could accurately
quantitate free and bound Neu5Gc. Initial base treatment of
samples to eliminate O-acetyl esters also ensured a simplified
profile for quantitation.
Using this improved method, we could accurately quantitate
bound and free Neu5Gc and N-acetylneuraminic acid (Neu5Ac)
in foods (examples of HPLC profiles are provided in Fig. 1). As
summarized in Table 1 and detailed in Table S3, foods from
mammals (cow, goat, sheep, pig, and bison) contain moderate to
2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1417508112

10

20 30 40
Time (min)

50

Fig. 1. Examples of DMB-HPLC chromatograms for quantification of Neu5Ac

(Ac) and Neu5Gc (Gc) content. Representative examples of results from different food groups are shown. Standard samples were run each day to account for slight variations in peak elution time.

high amounts of Neu5Gc. Three methods of cooking significantly altered neither Neu5Gc content nor the ratio of bound
and free forms. As predicted from prior work (33), poultry and
eggs do not contain Neu5Gc and fruits and vegetables do not
contain any sialic acids. Although the CMAH gene is present in
fish, none of the seafood sampled contains significant amounts of
Neu5Gc (with the exception of caviar). As shown in Table S3,
a great majority of the sialic acids in all samples are glycosidically
bound and not free, even after cooking. Overall, the highest levels
Samraj et al.

32
13
3253
922

0
0
0

0
0
0

0
0
445530
0
0

0
0
1329
0
0

Details within each category of foods are provided in Table S2.

of Neu5Gc among the red meats are in beef, which contains up to

231 g of Neu5Gc per gram of meat, and the lowest amounts were
seen in milk and milk products, with Neu5Gc levels ranging from 2
to 40 g/g. Beef also contains the highest percent Neu5Gc of total
sialic acid, and this Neu5Ac/Neu5Gc ratio may be relevant because
of likely competition of Neu5Ac with Neu5Gc for incorporation
into cells (34).
Systemic Inflammation Induced in Cmah/ Mice by Dietary Neu5Gc
and Circulating Anti-Neu5Gc Antibodies. We have previously shown

that anti-Neu5Gc antibodies can influence the growth of transplanted Neu5Gc-positive syngeneic tumors in Cmah/ mice by
enhancing chronic inflammation (16, 17). We now asked if oral
feeding of Neu5Gc could induce chronic inflammation in vivo,
and in a manner dependent on circulating anti-Neu5Gc antibodies. Cmah/ mice were fed a Neu5Gc-free diet or with the
same diet supplemented with Neu5Gc-rich porcine submaxillary
mucin (PSM), which contains 79% (wt/wt) of bioavailable glycosidically bound Neu5Gc, adds only a minimal amount of
neutral sugars and amino acids to the diet, and can cause mouse
tissue incorporation of Neu5Gc over a period of weeks at levels
histologically similar to the levels seen in adult humans who have
eaten red meat for many years (22). The question to be addressed is whether the addition of anti-Neu5Gc antibodies to
such animals would generate systemic inflammation. Such an
anti-Neu5Gc antibody response can be induced by active immunization (26). However, we found that this response is highly
variable and difficult to control or manipulate. We therefore
challenged Neu5Gc-fed or -nonfed mice by injection of antiNeu5Gcrich polyclonal serum, or with a highly specific control
serum, prepared as previously described (16, 18).
Cmah/ mice were fed Neu5Gc-rich or control diets for 12 wk
and then injected with varying amounts of control or immune
sera calculated to achieve levels of anti-Neu5Gc antibodies in
the range found in humans (34). As shown in Fig. 2A, evidence of
systemic inflammation (elevated levels of peritoneal fluid IL-6
and serum acute-phase proteins, serum amyloid A protein, and
haptoglobin) was seen only with the combination of dietary
Neu5Gc plus infusion of anti-Neu5Gc antisera, and not with
control combinations. Furthermore, the levels of inflammatory
Samraj et al.

test our hypothesis that oral feeding of Neu5Gc can result in

metabolic incorporation and interact with circulating antiNeu5Gc antibodies, we used the human-like, Neu5Gc-deficient
Cmah/ mice bred into a C57BL/6 background, in which occasional male mice develop liver tumors (30). Two groups of
such mice were immunized with a Neu5Gc-containing immunogen (porcine RBC ghosts) and fed either a Neu5Gc-rich diet
(PSM, containing 0.25 mg of Neu5Gc per gram of chow) or
a Neu5Ac-rich diet containing edible birds nest (EBN; containing 0.25 mg of Neu5Ac per gram of chow). Sera from the two
groups showed comparable levels of anti-Neu5Gc antibodies
(Fig. 3A), and there were no significant abnormalities in hematology or blood chemistry values. Although immunohistological
examination of such mice showed Neu5Gc incorporation at
levels similar to those levels seen in tissues from red meat-eating
humans (22), precise chemical methods for comparing incorporation are not yet available. We therefore monitored serum
levels of the inflammatory cytokine IL-6, which is frequently
elevated in chronic liver disease and malignancy (35, 36). Significantly higher levels were seen in the Cmah/ mice fed PSM
for up to 30 wk (Fig. 3B). We also monitored the mice with
periodic MRI scans. Based on the first appearance of a visible
liver lesion by MRI at 40 wk (Fig. 3C), we chose to necropsy all
mice at 55 wk. Two of seven Cmah/ mice fed PSM had early
hepatocellular carcinomas (HCCs) (Fig. 3D), supporting our
hypothesis that interaction between anti-Neu5Gc antibodies and
Neu5Gc of dietary origin facilitates carcinoma progression.
To confirm and extend the findings, we set up a comparison of
Cmah/ and WT mice fed Neu5Gc-rich PSM, and also included
control immunizations. We chose human and chimpanzee RBC
ghosts as immunogens for this set of experiments because these
cell types are very similar, except for Neu5Gc in the chimpanzee.
This experimental condition also mimics human populations (25)

30

p<0.0001

20
10
0

-10

PSM
Diet Soy
Sera Control Control

60

PSM
Immune

p < 0.0001

40

20

15 g/ml 30 g/ml 45 g/ml

150
p=0.0015
100

50

15

p<0.0001

10

PSM
Diet Soy
Sera Control Control

PSM
Immune

PSM
Diet Soy
Sera Control Control

150
p = 0.0036
100

50

15 g/ml 30 g/ml 45 g/ml

25

PSM
Immune

p < 0.0001

20
15
10
5
0

15 g/ml 30 g/ml

45 g/ml

Fig. 2. Dietary Neu5Gc and anti-Neu5Gc antibody-dependent inflammation in

Neu5Gc-deficient mice. Cmah/ mice were fed Neu5Gc-free or Neu5Gc-rich diets
for 12 wk and then injected with control or anti-Neu5Gc immune sera as described in Materials and Methods. Sera and peritoneal lavage fluid were collected after 5 d of transfer of sera and analyzed for various inflammatory
markers. (A) Inflammatory responses with different combinations of sera and
diets. (n = 6 mice in each of the control groups and n = 12 mice in the test
group). The probability values are P < 0.001, P = 0.0015, and P < 0.001 for IL-6,
serum amyloid A (SAA), and haptoglobin (Hp), respectively. (B) Dose dependency
of the inflammatory response in Neu5Gc-fed mice injected with immune sera (n
= 6 mice in each group). The probability values are P < 0.001, P = 0.0036, and P <
0.001 for IL-6, SAA, and Hp, respectively. Tests of significance were carried out in
Prism version 6.0 using the Student t test.

PNAS Early Edition | 3 of 6

MEDICAL SCIENCES

29
14
25231
740

Neu5Gc-Rich Diet Promotes Hepatocellular Cancer Incidence in

Cmah/ Mice Only in the Presence of Anti-Neu5Gc Antibodies. To

Serum HP (g/ml)

0
12
24
44

Serum HP (g/ml)

0
2
1022
43

Serum SAA (g/ml)

Neu5Gc, %

Serum SAA (g/ml)

Dairy
Butter
Whole milk
Cows milk cheeses
Goats milk cheese
Red meats
Bison
Lamb
Beef
Pork
Poultry
Hen egg
Turkey
Chicken
Seafood
Fish
Shellfish
Caviar
Vegetables
Fruits

Neu5Gc content, g/g

Peritoneal Lavage
IL-6 ng/ml

Food sample

markers showed a dose dependency with the amount of antibody

injected (Fig. 2B). Similar studies were not done in Cmah+/+
mice, because high levels of endogenous Neu5Gc would neutralize any transferred antibodies (24).

Peritoneal Lavage
IL-6 ng/ml

Table 1. Summary of Neu5Gc content and percentage of

Neu5Gc (relative to total sialic acids) of various food groups

Fig. 3. Spontaneous appearance of HCCs dependent on dietary Neu5Gc and

anti-Neu5Gc antibodies. As described in Materials and Methods, Cmah/
mice immunized against Neu5Gc were fed a Neu5Gc (PSM)- or Neu5Ac (EBN)rich diet for a period of 55 wk (n = 7 in the PSM group and n = 6 in the EBN
group). (A) Anti-Neu5Gc antibody levels in Cmah/ mice at 2 wk postimmunization. (B) Significantly higher levels of IL-6 are seen in mice fed
a Neu5Gc-rich diet. Tests of significance were carried out in Prism version 6.0
using two-way ANOVA, and the probability value is P = 0.0080. (C, Upper) T1weighted MRI images postgadolinium injection demonstrating a focal contrast-enhancing lesion in the left lobe of the liver. (C, Lower) HCC tumor
margin (black arrows) in H&E staining of the Cmah/ mouse livers fed
a Neu5Gc-rich diet (PSM). (Scale bar: 100 m; Magnification: 100.) (D) Table
detailing genotype, immunization, and HCC rates.

in that the antibody response is polyclonal and variable. We also

extended the duration of follow-up to >80 wk, during which
variable numbers of mice in each group either died spontaneously or had to be euthanized for severe dermatitis or other
moribund states. There was no specific pattern to these losses,
and necropsies showed no liver tumors. All surviving mice were
finally killed at 85 wk. As expected, occasional mice in some
groups had hepatic adenomas or HCCs at this late time point
(incidence ranging from 09%). However, only those Cmah/
mice that were fed PSM and immunized with Neu5Gc-rich RBC
ghosts developed HCCs at a substantially higher rate (47%; i.e.,
the mice expected to have xenosialitis). This experiment was
repeated in a second cohort, with even clearer results (i.e., carcinomas occurred only in the Cmah/ mice fed PSM and immunized with Neu5Gc). The raw data from both experiments are
shown in Table S4, and the combined data are summarized in
Fig. 4A. Overall, in striking contrast to all of the other groups,
almost half of the Cmah/ mice fed PSM and immunized with
chimpanzee RBC ghosts developed HCCs (histological examples
are shown in Fig. 4B, including an example of lung metastasis).
Immunohistochemistry also demonstrated incorporation of
Neu5Gc into the tumors (Fig. 4C). Taken together, our data
show that the combination of dietary Neu5Gc and circulating
anti-Neu5Gc antibodies specifically enhances carcinoma rates
only in Neu5Gc-deficient animals (P = 0.0043), in an organ that
is already prone to a baseline rate of tumorigenesis. This observation is very similar to the situation of human red meat
eaters, who show an increased incidence of carcinomas in tissues,
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1417508112

such as the colonic epithelium, that are already prone to

a baseline rate of cancer incidence. To our knowledge, this
model is the first example in which incorporation of a dietary
molecule combined with antibodies against the molecule is both
necessary and sufficient to induce spontaneous carcinomas.
Discussion
We have demonstrated here that antibodies directed against the
nonhuman sialic acid Neu5Gc can interact with metabolically incorporated Neu5Gc derived from oral intake and promote
inflammation in a dose-dependent manner in the human-like
Neu5Gc-deficient Cmah/ mouse model. Furthermore, long-term
exposure to such inflammation promotes carcinoma incidence in
a target organ where Neu5Gc can accumulate; in the case of mice,
it is detected in the liver, and in humans, it is detected more
prominently in the colon, prostate, and ovary. Given that glycosidically bound Neu5Gc is relevant for incorporation from dietary
sources, we have also quantified free and bound Neu5Gc content
in various food products, affirming that foods of mammalian origin
are the primary sources.
Chronic inflammation in the context of carcinogenesis and tumor
progression has been studied extensively. Although the adaptive
immune system can restrict cancer development, it can also paradoxically assist the tumor by promoting chronic inflammation via
antibodies directed against tumor cell epitopes (37, 38). Known
mechanisms include deposition of sublytic levels of complement
with activation of the PI3K/AKT survival pathway (39) and/or
Fc receptor-dependent polarization of F4/80+/CD11b+ tumorassociated macrophages into tumor-promoting M2 phenotypes,
thereby suppressing tumor infiltration by activated CD8+ lymphocytes
(40). Furthermore, it is possible that intratumoral antibody

p= 0.0043

Fig. 4. Neu5Gc-rich diet promotes hepatocellular cancer only in Neu5Gcdeficient mice immunized against Neu5Gc. As described in Materials and
Methods, WT or Cmah/ mice immunized with Neu5Gc- or Neu5Ac-rich RBC
ghosts (chimpanzee or human, respectively) were fed a Neu5Gc-rich (PSM) diet
for a period of 8085 wk before necropsy. Combined data from two experiments are shown. (A) HCC incidence in four groups of mice is compared: Cmah/
mice immunized with Neu5Gc, Cmah/ mice immunized with Neu5Ac, WT mice
immunized with Neu5Gc, and WT mice immunized with Neu5Ac. A markedly
higher rate of HCC was seen in the Neu5Gc-fed and Neu5Gc-immunized Cmah/
mice compared with the control groups (the raw data are provided in Table S4).
Tests of significance were carried out using the Fisher exact probability test. The
two-tailed probability value for the test group (Cmah/ Neu5Gc) compared with
the control groups is P = 0.0043. (B) Representative examples of HCC in Cmah/
mouse liver are shown, with black arrows pointing to the tumor interface. An
example of lung metastasis is also shown. (Scale bar: 100 m; Magnification:
40.) (C) Detection of Neu5Gc in tumors by immunohistochemistry. Representative examples of Cmah/ mouse liver are seen below. Positive staining with
anti-Neu5Gc IgY (Left) and competitive inhibition with chimpanzee serum as the
negative control (Right) are shown. (Scale bar: 100 m; Magnification: 200.)

Samraj et al.

Materials and Methods

Food Collection and DMB-HPLC Analysis. Sixty-two different types of food
products (raw meat and three types of cooked meats, including baked,
boiled, and fried foods, were sampled) were purchased from multiple local
supermarkets, lyophilized, and pulverized. Samples were treated with 0.1 M
NaOH at 37 C for 30 min for the removal of O-acetyl groups and then
neutralized. Acid hydrolysis of sialic acids from the underlying glycoconjugate was performed with 2 M acetic acid at 80 C for 3 h. The samples
were cooled to room temperature and centrifuged at 14,000 g for 10 min,
and the supernatant was filtered through a Millipore Ultra Centrifugal Filter. Aliquots of nonhydrolyzed samples were adjusted to 2 M acetic acid but
not heated. The samples were then incubated with the two different derivatizing conditions (below) and analyzed by HPLC as described below.
DMB Derivatization Conditions. The DMB reagent, as per the conventional
method (31, 46), was made with the following recipe and hydrolyzing conditions: 14 mM DMB, 18 mM sodium hydrosulfite, 0.75 M 2-mercaptoethanol, and 1.6 M acetic acid, and it was incubated at 50 C for 2.5 h. The DMB
reagent used for the quantification of free sialic acids was prepared as follows (32): 14 mM DMB, 18 mM sodium hydrosulfite, 1 M 2-mercaptoethanol,
and 40 mM trifluoroacetic acid, and it was incubated at 4 C for 48 h.
HPLC Analysis. The DMB-derivatized samples were analyzed on a Dionex
Ultra3000 HPLC System using a Phenomenex Gemini 5 C18 250 4.6-mm
HPLC column at room temperature. The fluorescence was detected at 448
nm using excitation at 373 nm. For the separation of sialic acids that might
coelute, for example, 3-deoxy-D-manno-octulosonic acid and 3-deoxy-Dmanno-octulosonic acid, an isocratic solvent composition of 88% water, 7%
methanol, and 5% acetonitrile was used. The data collection time was expanded to 90 min. Standard samples were run each day to account for slight
variations in elution time.
Mice and Chow. Cmah/ mice were bred in a congenic C57BL/6 background
and maintained in the University of California, San Diego vivarium according
to Institutional Review Board guidelines for the care and use of laboratory
animals. Cmah/ mice were maintained on a Neu5Gc-deficient soy-based chow

Samraj et al.

(110951 for adults and 110751 for pregnancy/weaning; Dyets, Inc.), a Neu5Gcrich soy-based chow (custom order containing 0.25 mg of Neu5Gc per gram of
chow; Dyets, Inc.) made by adding purified PSM, or a Neu5Ac-rich soy-based
chow (custom order containing 0.25 mg of Neu5Ac per gram of chow; Dyets,
Inc.) made by adding EBN. For the preparation of PSM, cryoground porcine
submaxillary glands (Pel-Freez Biologicals) were homogenized overnight in
5 vol of water and centrifuged at 8,000 g for 15 min, and the supernatant
was filtered through glass wool. The mucin was precipitated by gradual
acidification (to pH 3.5) at 4 C and left to settle overnight. The supernatant
was removed by siphoning, and the precipitated mucin was centrifuged at
400 g for 15 min, washed with water, and centrifuged again. Mucin pellets
were neutralized to pH 8.0 and dialyzed using a 10,000 molecular weight
cutoff membrane (Spectrum Labs) against 20 vol of water, with at least five
volume changes. This final preparation was lyophilized, and its Neu5Gc
content was quantified by DMB-HPLC. EBN was purchased from Golden Nest,
Inc. and lyophilized to remove excess moisture. The dry EBN was pulverized
into a fine powder, and acid hydrolysis for DMB derivatization was performed
as mentioned above. It must be noted that the addition of PSM or EBN does not
significantly increase the caloric content of the chow.
Preparation of RBC Membranes and Polyclonal Anti-Neu5GcRich Immune
and Control Sera. Chimpanzee and human RBC lysis was performed using
30 vol of ice-cold lysis buffer containing 10 mM TrisHCl and 1 mM EDTA (pH
7.4) followed by centrifugation at 15,000 g for 15 min at 4 C. Pelleted
membranes were washed in the same buffer until colorless. Protein quantification was performed, and Cmah/ mice were immunized with 100 L of
2 mg/mL RBC membrane ghosts (mixed with an equal volume of Freunds
complete adjuvant) per mouse by i.p. injection. Two booster doses using
Freunds incomplete adjuvant with the same amount of immunogen were
given 1 wk apart. Two weeks after the second booster dose, serum was
collected for ELISA analysis of anti-Neu5Gc response as described below.
Positive sera were pooled, and nonspecific anti-RBC reactivity was removed
by repeated adsorption against human RBCs. Adsorption was performed
using 100 L of packed washed human RBCs incubated with pooled immune
or control sera at 4 C for 2 h, and the RBCs were subsequently removed by
centrifugation. The adsorption was repeated three times. The pooled serum
from the human RBC-immunized mice was processed in exactly the same
way. The mouse anti-Neu5Gc IgG was quantitated using the ELISA method
described below. Based on the above method, these antibodies should
comprise the primary difference between the immune and control sera.
ELISAs for Detection of Anti-Neu5Gc Antibodies in Mice. Microtiter plate (Costar
9018) wells were coated with Neu5Gc-rich bovine submaxillary mucin in the
amount of 1 g per well in 50 mM sodium carbonate-bicarbonate buffer (pH
9.5) at 4 C overnight. After washing with PBS (pH 7.4) and blocking with PBS
containing 1% Neu5Gc-free chicken ovalbumin for 1 h at room temperature,
triplicate wells were incubated with 1:100 dilutions of mouse serum in PBS
containing 1% chicken ovalbumin at room temperature for 2 h. Wells were
washed five times with PBS containing 0.05% Tween 20 (Fisher-Scientific) and
incubated with HRP-conjugated goat anti-mouse IgG (Jackson ImmunoResearch
Laboratories), diluted in PBS at 1:7,000, at room temperature for 1 h. AntiNeu5Gc IgG was quantified using a standard curve of normal mouse IgG coated
to the wells under the same general conditions. To confirm the specificity of
anti-Neu5Gc antibodies, mild periodate treatment was used selectively to cleave
the Sia side-chain and eliminate specific reactivity, as described previously (47).
Measurement of Peritoneal Lavage Fluid IL-6, Serum Amyloid A Protein, and
Haptoglobin. Commercially available kits were used for the measurement of
inflammatory markers, peritoneal lavage fluid IL-6 (no. DY406, Mouse Il-6
DuoSet; R&D Systems), serum haptoglobin (no. 2410-1; Life Diagnostics, Inc.),
and serum amyloid protein A (no. TP 802M; Tridelta) according to the
manufacturers instructions.
Neu5Gc Detection by Immunohistochemistry. Tissues from Cmah/ mice fed or
not fed PSM were flash-frozen in optimum cutting temperature compound
(Sakura), and frozen sections were air-dried overnight at room temperature and
then immersed in PBS containing 0.1% Tween (PBST) wash buffer. The sections on
the slides were treated to eliminate endogenous biotin and endogenous
peroxidase, and postfixed in 10% neutral buffered formalin (Fisher), washing
between each step. Positive controls on separate slides included sections from
WT mouse liver, and negative controls included sections from Cmah/ mouse
livers. The negative controls used competitive inhibition of anti-Neu5Gc
binding, by incubating 20% chimpanzee serum with the anti-Neu5Gc antibody for 1 h at 4 C before overlaying on serial frozen sections. All slides
were incubated in a humid chamber for at least 1 h at room temperature

PNAS Early Edition | 5 of 6

MEDICAL SCIENCES

deposition accelerates chronic inflammation by inducing pathways involving COX-2 (41, 42) and NF-B (43). Release of reactive oxygen species by recruited phagocytic cells could also
cause DNA damage, facilitating carcinogenesis (44). On the
other hand, there are many dramatic examples of the successful
therapy of cancer with monoclonal antibodies (45). In our own
recent work, we showed that the range of antitumor antibody
levels that bridge the spectrum from tumor activation to inhibition can be remarkably narrow (18).
There are limitations in directly comparing this mouse study
and the human situation. Unlike the case in humans, there was
a single dietary source of Neu5Gc that was not varied in intake
and the antibody production was not sustained throughout the
lifetime of the animal. Also, the target organ for the adenomato-carcinoma sequence was the liver, not the colon. The obvious
question arising from such experimental mouse studies is
whether circulating anti-Neu5Gc antibodies correlate with cancer risk in human population studies. We are currently exploring
this possibility, but we are also aware of the numerous complicating variables in humans, such as the varying amount and unknown bioavailability of Neu5Gc from dietary red meat, the
amount of Neu5Gc tissue loading in benign and malignant tissues, the complex and variable polyclonal antibody profiles of
individual humans, the skewed distribution of antibody levels
within populations, the possibility that very high antibody levels
can inhibit tumor progression (18), and the variability of concurrent inflammatory conditions.
Further work is also necessary to determine the exact pathways
by which Neu5Gc is taken up into tissues, and potential approaches
to prevent or eliminate such incorporation are being explored. The
current findings indicate that this unique example of xenosialitis
driven by a dietary glycan is pathologically relevant and may be
involved in inflammation-driven diseases that are known to be associated with red meat consumption, including carcinomas.

with anti-Neu5Gc IgY or with the control IgY (Sialix, Inc.) at 5 g/mL
[1:2,000 diluted in blocking buffer, 0.5% cold water fish skin gelatin
(Sigma) in PBST]. Slides were then overlaid successively with the biotinylated donkey anti-chicken IgY (1:500; Jackson ImmunoResearch) and
with HRP-streptavidin (1:500; Jackson ImmunoResearch) for 30 min each.
Substrate color development used 3-amino-9-ethylcarbazole (Vector Laboratories AEC substrate kit), and the nuclei were counterstained with
Mayers hematoxylin (Sigma). Slides were coverslipped using Aqueous
VectaMount (Vector Laboratories) for viewing, and digital photomicrography was performed using a Keyence B6000 microscope.

Statistical Analyses. Statistical analyses were performed on GraphPad Prism

version 6.0 or Microsoft Office Excel. Relevant tests of significance and the
statistical tests used are further detailed in the figure legends.

1. Snowdon DA (1988) Animal product consumption and mortality because of all causes
combined, coronary heart disease, stroke, diabetes, and cancer in Seventh-day Adventists. Am J Clin Nutr 48(3, Suppl):739748.
2. Song Y, Manson JE, Buring JE, Liu S (2004) A prospective study of red meat consumption and type 2 diabetes in middle-aged and elderly women: The womens
health study. Diabetes Care 27(9):21082115.
3. Sinha R, Cross AJ, Graubard BI, Leitzmann MF, Schatzkin A (2009) Meat intake and
mortality: A prospective study of over half a million people. Arch Intern Med 169(6):
562571.
4. Pan A, et al. (2012) Red meat consumption and mortality: Results from 2 prospective
cohort studies. Arch Intern Med 172(7):555563.
5. Glass CK, Olefsky JM (2012) Inflammation and lipid signaling in the etiology of insulin
resistance. Cell Metab 15(5):635645.
6. Rocha VZ, Libby P (2009) Obesity, inflammation, and atherosclerosis. Nat Rev Cardiol
6(6):399409.
7. Esmaillzadeh A, et al. (2007) Dietary patterns and markers of systemic inflammation
among Iranian women. J Nutr 137(4):992998.
8. Fraser GE (1999) Associations between diet and cancer, ischemic heart disease, and allcause mortality in non-Hispanic white California Seventh-day Adventists. Am J Clin
Nutr 70(3, Suppl):532S538S.
9. Key TJ, et al. (1999) Mortality in vegetarians and nonvegetarians: Detailed findings
from a collaborative analysis of 5 prospective studies. Am J Clin Nutr 70(3, Suppl):
516S524S.
10. Singh PN, Sabat J, Fraser GE (2003) Does low meat consumption increase life expectancy in humans? Am J Clin Nutr 78(3, Suppl):526S532S.
11. Wiseman M (2008) The second World Cancer Research Fund/American Institute for
Cancer Research expert report. Food, nutrition, physical activity, and the prevention
of cancer: A global perspective. Proc Nutr Soc 67(3):253256.
12. Fearon ER (2011) Molecular genetics of colorectal cancer. Annu Rev Pathol 6:479507.
13. Cross AJ, et al. (2010) A large prospective study of meat consumption and colorectal
cancer risk: An investigation of potential mechanisms underlying this association.
Cancer Res 70(6):24062414.
14. Sugimura T, Wakabayashi K, Nakagama H, Nagao M (2004) Heterocyclic amines:
Mutagens/carcinogens produced during cooking of meat and fish. Cancer Sci 95(4):
290299.
15. Stavric B (1994) Biological significance of trace levels of mutagenic heterocyclic aromatic amines in human diet: A critical review. Food Chem Toxicol 32(10):977994.
16. Hedlund M, Padler-Karavani V, Varki NM, Varki A (2008) Evidence for a human-specific mechanism for diet and antibody-mediated inflammation in carcinoma progression. Proc Natl Acad Sci USA 105(48):1893618941.
17. Padler-Karavani V, et al. (2011) Human xeno-autoantibodies against a non-human
sialic acid serve as novel serum biomarkers and immunotherapeutics in cancer. Cancer
Res 71(9):33523363.
18. Pearce OM, et al. (2014) Inverse hormesis of cancer growth mediated by narrow
ranges of tumor-directed antibodies. Proc Natl Acad Sci USA 111(16):59986003.
19. Samraj AN, Lubli H, Varki N, Varki A (2014) Involvement of a non-human sialic Acid
in human cancer. Front Oncol 4:33.
20. Varki A (2010) Colloquium paper: Uniquely human evolution of sialic acid genetics
and biology. Proc Natl Acad Sci USA 107(Suppl 2):89398946.
21. Bergfeld AK, Pearce OM, Diaz SL, Pham T, Varki A (2012) Metabolism of vertebrate
amino sugars with N-glycolyl groups: Elucidating the intracellular fate of the nonhuman sialic acid N-glycolylneuraminic acid. J Biol Chem 287(34):2886528881.
22. Banda K, Gregg CJ, Chow R, Varki NM, Varki A (2012) Metabolism of vertebrate
amino sugars with N-glycolyl groups: Mechanisms underlying gastrointestinal incorporation of the non-human sialic acid xeno-autoantigen N-glycolylneuraminic
acid. J Biol Chem 287(34):2885228864.
23. Tangvoranuntakul P, et al. (2003) Human uptake and incorporation of an immunogenic nonhuman dietary sialic acid. Proc Natl Acad Sci USA 100(21):1204512050.
24. Hedlund M, et al. (2007) N-glycolylneuraminic acid deficiency in mice: Implications for
human biology and evolution. Mol Cell Biol 27(12):43404346.

25. Padler-Karavani V, et al. (2008) Diversity in specificity, abundance, and composition of

anti-Neu5Gc antibodies in normal humans: Potential implications for disease. Glycobiology 18(10):818830.
26. Taylor RE, et al. (2010) Novel mechanism for the generation of human xeno-autoantibodies against the nonhuman sialic acid N-glycolylneuraminic acid. J Exp Med
207(8):16371646.
27. Spichtig V, Michaud J, Austin S (2010) Determination of sialic acids in milks and milkbased products. Anal Biochem 405(1):2840.
28. Chen Y, Pan L, Liu N, Troy FA, Wang B (2014) LC-MS/MS quantification of N-acetylneuraminic acid, N-glycolylneuraminic acid and ketodeoxynonulosonic acid levels in the urine
and potential relationship with dietary sialic acid intake and disease in 3- to 5-year-old
children. Br J Nutr 111(2):332341.
29. Li H, Fan X (2014) Quantitative analysis of sialic acids in Chinese conventional foods by
HPLC-FLD. Open J Prev Med 4:57.
30. Nakamura K, Kuramoto K, Shibasaki K, Shumiya S, Ohtsubo K (1992) [Age-related
incidence of spontaneous tumors in SPF C57BL/6 and BDF1 mice]. Jikken Dobutsu
41(3):279285. Japanese.
31. Hara S, et al. (1989) Determination of mono-O-acetylated N-acetylneuraminic acids in
human and rat sera by fluorometric high-performance liquid chromatography. Anal
Biochem 179(1):162166.
32. Inoue S, Inoue Y (2003) Ultrasensitive analysis of sialic acids and oligo/polysialic acids
by fluorometric high-performance liquid chromatography. Methods Enzymol 362:
543560.
33. Schauer R, Srinivasan GV, Coddeville B, Zanetta JP, Gurardel Y (2009) Low incidence
of N-glycolylneuraminic acid in birds and reptiles and its absence in the platypus.
Carbohydr Res 344(12):14941500.
34. Ghaderi D, Taylor RE, Padler-Karavani V, Diaz S, Varki A (2010) Implications of the
presence of N-glycolylneuraminic acid in recombinant therapeutic glycoproteins. Nat
Biotechnol 28(8):863867.
35. Jin X, et al. (2006) Paradoxical effects of short- and long-term interleukin-6 exposure
on liver injury and repair. Hepatology 43(3):474484.
36. Naugler WE, et al. (2007) Gender disparity in liver cancer due to sex differences in
MyD88-dependent IL-6 production. Science 317(5834):121124.
37. Balkwill F, Charles KA, Mantovani A (2005) Smoldering and polarized inflammation in
the initiation and promotion of malignant disease. Cancer Cell 7(3):211217.
38. Tan TT, Coussens LM (2007) Humoral immunity, inflammation and cancer. Curr Opin
Immunol 19(2):209216.
39. Wu X, Ragupathi G, Panageas K, Hong F, Livingston PO (2013) Accelerated tumor
growth mediated by sublytic levels of antibody-induced complement activation is
associated with activation of the PI3K/AKT survival pathway. Clin Cancer Res 19(17):
47284739.
40. Andreu P, et al. (2010) FcRgamma activation regulates inflammation-associated
squamous carcinogenesis. Cancer Cell 17(2):121134.
41. Mann JR, Backlund MG, DuBois RN (2005) Mechanisms of disease: Inflammatory
mediators and cancer prevention. Nat Clin Pract Oncol 2(4):202210.
42. Rothwell PM, et al. (2011) Effect of daily aspirin on long-term risk of death due to
cancer: Analysis of individual patient data from randomised trials. Lancet 377(9759):
3141.
43. Grivennikov SI, Greten FR, Karin M (2010) Immunity, inflammation, and cancer. Cell
140(6):883899.
44. Loft S, Poulsen HE (1996) Cancer risk and oxidative DNA damage in man. J Mol Med
(Berl) 74(6):297312.
45. Sliwkowski MX, Mellman I (2013) Antibody therapeutics in cancer. Science 341(6151):
11921198.
46. Manzi AE, Diaz S, Varki A (1990) High-pressure liquid chromatography of sialic acids
on a pellicular resin anion-exchange column with pulsed amperometric detection: A
comparison with six other systems. Anal Biochem 188(1):2032.
47. Padler-Karavani V, et al. (2013) A simple method for assessment of human antiNeu5Gc antibodies applied to Kawasaki disease. PLoS ONE 8(3):e58443.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1417508112

ACKNOWLEDGMENTS. We thank Profs. Walter Willett (Harvard University)

and Kay-Tee Khaw (Cambridge University) for helpful discussions, and
Michelle Y. Chung, Pratyash Srivastava, and Siavash Assar for help with
some of the experiments. This work was supported by a Samuel and Ruth
Engelberg Fellowship from the Cancer Research Institute (to O.M.T.P.), the
Swiss National Science Foundation (H.L.), the Ellison Medical Foundation
(A.V.), and NIH Grant R01CA38701 (to A.V.).

Samraj et al.